Articles Journal of Attention Disorders Volume 11 Number 2 September 2007 106-113 © 2007 Sage Publications 10.1177/1087054707300094 http://jad.sagepub.com hosted at Evaluating the Evidence For and Against http://online.sagepub.com the Overdiagnosis of ADHD Mark J. Sciutto Miriam Eisenberg Muhlenberg College, Pennsylvania Objective: According to the DSM-IV TR, approximately 3 to 7% of school-age children meet the criteria for ADHD. However, there is a common conception that ADHD is overdiagnosed. The purpose of this article is to evaluate the evidence for and against overdiagnosis. Method: Recent prevalence studies and research on factors affecting diagnostic accuracy were reviewed. For ADHD to be overdiagnosed, the rate of false positives (i.e., children inappropriately diagnosed with ADHD) must substantially exceed the number of false negatives (children with ADHD who are not identified or diagnosed). Results and Conclusion: Based on the review of prevalence studies and research on the diagnostic process, there does not appear to be sufficient justifi- cation for the conclusion that ADHD is systematically overdiagnosed. Yet, this conclusion is generally not reflected in public perceptions or media coverage of ADHD. Potential explanations for the persistence of the belief in the overdiagnosis of ADHD are offered. (J. of Att. Dis. 2007; 11(2) 106-113) Keywords: ADHD; assessment; misdiagnosis; overdiagnosis; diagnosis; prevalence DHD is one of the most commonly diagnosed child- poll in 2002 found that 76% of respondents believed that Ahood disorders (Barkley, 2005). According to the ADHD is overdiagnosed (Do You Think ADHD, 2002). DSM-IV TR, approximately 3 to 7% of school-age When people speak of ADHD being overdiagnosed, children meet the criteria for ADHD (American Psychiatric they are typically referring to children who are diag- Association [APA], 2000). However, there is a common nosed with ADHD but should not be (i.e., false posi- conception that ADHD is vastly overdiagnosed, with some tives). For example, Kissinger (1998) introduces the suggesting that it is the diagnosis du jour (Bogas, 1997) topic of overdiagnosis by describing a child who was ini- and that it may be a desirable diagnosis for some parents tially believed to have ADHD because he was persis- (Smelter & Rasch, 1996). This theme has been clearly tently inattentive even during “a popular study unit on reflected in the coverage of ADHD in popular press and cowboys” (p. 1). Only after additional information was gath- mass media sources (e.g., Brazelton & Sparrow, 2003; ered did the origins of his problems become apparent—he Haber, 2000; Kissinger, 1998; Parenting with Pills, 2004). was experiencing traumatic stress as a result of being sexu- For example, after an episode on parenting hyperactive ally abused by a man who had worn cowboy boots. Clearly, children, talk show host Dr. Phil McGraw claimed that this child should not have been diagnosed with ADHD ADHD is “so overdiagnosed” (Parenting with Pills, 2004). and cases like this might be expected if ADHD were in The issue of overdiagnosis has also been addressed by fact overdiagnosed. However, addressing the question of prominent politicians like Hillary Clinton. In a press con- overdiagnosis is very complex and requires looking for ference in March of 2000, she voiced concern about over- evidence beyond only the cases in which a child is diagnosis by suggesting that physicians might be too quick wrongly given a diagnosis of ADHD. to diagnose children whose problems may be simply nor- mal characteristics of childhood and adolescence (Vatz & Authors’ Note: Address correspondence to Mark J. Sciutto, PhD, Weinberg, 2001). These examples appear to be consistent Department of Psychology, Muhlenberg College, Allentown, PA, with public sentiment about overdiagnosis. A CNN online 18104; [email protected] 106 Sciutto, Eisenberg / Is ADHD Overdiagnosed? 107 Implicit in the overdiagnosis question is that there is a criteria) and coded them for potential moderating vari- clearly definable level of how many children should be ables such as the number and type of informants, sample diagnosed with ADHD. If more children are receiving a characteristics, and the method of assessment. We diagnosis of ADHD than should be, then it would seem included any study that explicitly provided an estimate valid to conclude that ADHD is indeed overdiagnosed. of the prevalence of ADHD among school-age children However, the presence of false positives alone does not in the United States. Studies were located through a indicate overdiagnosis. With any diagnostic system that is PsychINFO search and through existing reviews of not perfectly reliable, there will also be children who war- prevalence studies (Barkley, 2005; Faraone, Sergeant, rant the diagnosis, but go unidentified or undiagnosed Gillberg, & Biederman, 2003). A summary of these stud- (i.e., false negatives). For ADHD to be overdiagnosed, the ies is presented in Table 1. overall number of false positives must substantially Results indicated that the prevalence of ADHD is exceed the overall number of false negatives. often above the 3 to 7% range cited by the DSM-IV. The most direct test of the overdiagnosis question Consistent with the DSM-IV TR, the male to female would involve a nationally representative study in which ratio in these studies ranged from 1.58 to 6.36 (M = 2.73, diagnoses based on standardized, multimethod assess- SD = 1.12). The overall prevalence of ADHD varied con- ments were compared to actual diagnoses. If the number siderably depending on the assessment method and age. of children who have been actually diagnosed with For instance, the Center for Disease Control study docu- ADHD exceeds the number that would meet the criteria mented the considerable variability in prevalence across based on a multimethod assessment, then there would be age, gender, and geographic location (Center for Disease direct evidence of overdiagnosis. In other words, the issue Control and Prevention, 2005). The prevalence estimates of overdiagnosis will be resolved only when we compare were within the 3 to 7% range for younger children (ages the diagnoses being given in actual practice to the diag- 4 to 8) and slightly above the 3 to 7% range for older noses that should have been given based on standardized children (ages 9 to 17). In the 14 studies we reviewed, comprehensive assessments. Unfortunately, this type of methodological characteristics also appeared to affect study has not been conducted to date. Alternatively, there prevalence estimates. Specifically, the prevalence of may be indirect evidence for overdiagnosis if there are sys- ADHD was more likely to be lower and within the 3 to tematic inaccuracies in either (a) the accepted prevalence 7% range when studies used random sampling and larger rates or (b) in the diagnostic process. In this paper, we samples. In contrast, several studies documented preva- examine the evidence for and against the overdiagnosis lence rates that were substantially higher than the question in each of these areas. expected 3 to 7%. Typically, these prevalence rates used nonrandom samples, used screening measures, and/or Prevalence Rates relied heavily on a single information source. Because the prevalence estimates are so clearly linked to method- ology, it is unclear whether the 3 to 7% significantly With regard to prevalence rates, it would be reason- overestimates or underestimates the number of children able to conclude that ADHD is overdiagnosed if there is actually being diagnosed. Despite being frequently cited systematic evidence that the DSM-IV TR estimate of 3 in discussions of overdiagnosis, prevalence rates provide to 7% is either too high (i.e., many children falsely diag- relatively little interpretable evidence because of the sub- nosed with ADHD are included in the 3 to 7%) or too stantial variations in methodology (Cohen, Riccio, & low (i.e., the 3 to 7% underestimates the number of Gonzalez, 1994). children actually being diagnosed). Again, without a direct comparative study, as described above, prevalence rates alone provide little insight into this issue, but may The Diagnostic Process reflect trends consistent with systematic biases in the diagnosis of ADHD. For instance, if most prevalence Conceptually, looking at prevalence studies alone reveals studies exceed the 3 to 7% range, then it would be a rea- very little about overdiagnosis per se. Overdiagnosis, by sonable assumption that the DSM-IV estimate is too low. definition, implies a comparison to a reference point. When Likewise, if the existing prevalence studies fall within people speak of ADHD being overdiagnosed, they may not the 3 to 7% range, then it is less plausible to argue for be referring to the 3 to 7% prevalence estimate but to the systematic overdiagnosis. To evaluate the validity of the frequency of false positives—more children are being diag- DSM-IV prevalence estimates, we reviewed 14 recent nosed than should be. As mentioned previously, the pres- prevalence studies (since the publication of the DSM-IV ence of false positives by itself does not indicate 108 Journal of Attention Disorders Table 1 Recent Studies Estimating the Prevalence of ADHD in U.S. Samples Prevalence Estimate (%) Author Sample Size Random Sample Assessment Ages (Grades) Overall Male Female Barbaresi et al. (2002) 5,718 N A(medical, 5-19 7.5a (Definite) / 10.8 / 13.3 3.9 / 5.1 school) 15.9 (Definite or probable) Carlson et al. (1997) 2,984 N BRS-T (1-5) 18.9 25.5 11.6 Costello et al. (2003) 6,674 Y I-P 9-16 4.1b 7.0 1.1 Cuffe et al. (2005) 10,255 N BRS-P 4-17 3.3c 4.8 1.8 Gaub & Carlson (1997a) 2,744 N BRS-T (K-5) 8.1 — — Gimple & Kuhn (1998) 253 N PR 2-6 9.5 11.8 6.8 LeFever et al. (1999) 29,734 N A (2-5) 8.2-9.9 14.1 c 5.5 c LeFever et al.